Random Forest / XGBOOST

[Brain]

Hello,
I want to use XGBOOST instead of Randsom Forest in this workflow i found in KNIME.
I cant execute the metanode : Calculate feature importance.
Is it a problem of output ?
Thanks

[steffen_KNIME]

Dear @Brain,

please provide the failing workflow and/or describe clearly what the error is. I do not see any error in the screenshot.

Thanks
Steffen

[Brain]

I can not execute the metnaode

[Daniel_Weikert]

if there is an error then you can find that when you right click an open/expand the metanode
br

[Brain]

JKISeason2-9-tark.knwf @tark (131.2 KB)
Starting from this workflow. I want to use XGBOOST but i don’t know how to configure math formula in the metanode to obtain the same result.
Thanks for your help/

[tark]

Dear @Brain
Attribute statistics(splits and candidates), which are output from random forest node, are used for calculating the feature importance, but XGboost node doesn’t ouput them. So I don’t think this workflow can be used as is with Xgboost. Instead, Global feature importance may help you. Please see the following thread.

Calculation of feature relative importance from XG Boost Regression Tree

Hi, Is there any way to calculate the feature relative importance from the XG Boost Learner (Regression) in knime? Thanks.

[Brain]

@tark
I don’t find where we can change the tittle
Example (RF Four Classes) instead of (RF Two Classes)

Thanks

[tark]

Dear @Brain ,
In brief, open the component - bar chart - general plot options - title
After executing the component, you can change the configure of the bar chart.